Choosing Data Resolution and Platforms for Strategy Simulation
Summary
The document discusses ways to simulate a long-only S&P 500 strategy, including using hosted research platforms or building a local setup. It names Quantopian and QuantConnect as examples of online environments that support programmatic strategy development and historical research, with differences in supported languages, asset coverage, and available data resolution. The cited platform capabilities reflect the period and context of the original discussion, rather than a current platform comparison.
Data frequency should follow the questions a strategy is intended to answer. Daily observations can support some analyses, while minute, second, or tick-level records may reveal other behavior; finer resolution is not automatically necessary. The responses describe platform access to historical S&P data and varying frequencies, but do not give a universal recommendation, compare data quality, or specify a local data vendor. The user must match data needs to strategy behavior and available research tools.
Key ideas
- Select historical data frequency according to the behavior and questions the strategy is meant to study.
- Daily, minute, second, and tick data can reveal different aspects of trading behavior.
- Hosted research platforms can provide strategy coding environments and historical market data.
- Platform language support, asset coverage, and resolution differ, and cited offerings may change over time.
Tags
Full text
# Simulating trading strategies # Simulating trading strategies I want to simulate real time trading strategies. For simplicity, let's say I only want to simulate a long-only portfolio on S&P500. I have a couple of questions: - Is there a place online where you can simulate strategies programmatically? - Suppose that I want to build a simulator on my own machine. With historical data, what sort of frequency should I look at for collecting this data. Should I be looking at minute-by-minute data? On the second? Subsecond? Also, is there a standard place to acquire this old S&P Data? Disclaimer: Yes, I realize this is a broad question, and I appreciate helpful answers. I'm new to this. ## Answer by Theodore (score 6) https://quant.stackexchange.com/a/41317 > Is there a place online where you can simulate strategies programmatically? Your best choice is most likely a service such as Quantopian or QuantConnect. Quantopian provides equity and futures data and allows you to program trading strategies in Python, run risk management analysis and backtests. The latter option, QuantConnect, has support for Python as well as C# and F# in case you're more comfortable in either of those languages. QuantConnect also has higher resolution data for several more asset classes. I do not use either platforms as I prefer to be using a proprietary setup, plus most of what I do is in C++ and there aren't any online services for that regardless. I will not go into setting up a local environment for developing trading algorithms as you specifically asked for "a place online" and it's a loaded question. Regarding data resolution: QuantConnect allows you to specify what resolution of data you wish to use in your strategy. Daily, hour, minute and second data is provided by QuantConnect (as well as fundamental data). Same case for Quantopian but there the smallest resolution you will have access to is minute data. Regarding research frequency: This question is a bit broad but I'll try to cover what you're asking. It depends completely on what you are looking for. There are things you cannot find from daily data that you can find by looking at tick data and vice versa. As far as old S&P data goes, as stated above both Quantopian and QuantConnect provide data for researching purposes. (The research environments are effectively a Jupyter Notebook, and the language used for that is Python). This question was a bit broad but I hope I answered what you were looking for.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.